Copyright Ó 2008 by the Genetics Society of America DOI: 10.1534/genetics.108.088492

Identification of Amino Acid Residues in the Catalytic Domain of RNase E Essential for Survival of : Functional Analysis of DNase I Subdomain

Eunkyoung Shin,*,1 Hayoung Go,*,1 Ji-Hyun Yeom,* Miae Won,† Jeehyeon Bae,† Seung Hyun Han,‡ Kook Han,§ Younghoon Lee,§ Nam-Chul Ha,** Christopher J. Moore,†† Bjo¨rn Sohlberg,†† Stanley N. Cohen††,‡‡ and Kangseok Lee*,2 *Department of Life Science, Chung-Ang University, Seoul 156-756, Korea, †Graduate School of Life Science and Biotechnology, Pochon CHA University, Seongnam 463-836, Korea, ‡Department of Oral Microbiology and Immunology, School of Dentistry, Seoul National University, Seoul 110-749, Korea, §Department of Chemistry and Center for Molecular Design and Synthesis, Korea Advanced Institute of Science and Technology, Daejeon 305-701, Korea, **National Research Laboratory of Defense Proteins, College of Pharmacy, Pusan National University, Busan 609-735, Korea and ††Department of Genetics and ‡‡Department of Medicine, Stanford University, Stanford, California 94305 Manuscript received February 25, 2008 Accepted for publication May 12, 2008

ABSTRACT RNase E is an essential Escherichia coli that plays a major role in the decay and processing of a large fraction of in the cell. To better understand the molecular mechanisms of RNase E action, we performed a genetic screen for amino acid substitutions in the catalytic domain of the protein (N-Rne) that knock down the ability of RNase E to support survival of E. coli. Comparative phy- logenetic analysis of RNase E homologs shows that wild-type residues at these mutated positions are nearly invariably conserved. Cells conditionally expressing these N-Rne mutants in the absence of wild-type RNase E show a decrease in copy number of plasmids regulated by the RNase E substrate RNA I, and accumulation of 5S ribosomal RNA, M1 RNA, and tRNAAsn precursors, as has been found in Rne-depleted cells, suggesting that the inability of these mutants to support cellular growth results from loss of ribonucleolytic activity. Purified mutant proteins containing an amino acid substitution in the DNase I subdomain, which is spatially distant from the catalytic site posited from crystallographic studies, showed defective binding to an RNase E substrate, p23 RNA, but still retained RNA cleavage activity—implicating a previously unidentified structural motif in the DNase I subdomain in the binding of RNase E to targeted RNA molecules, demonstrating the role of the DNase I domain in RNase E activity.

MONG the many factors involved in the degradation 1992). The catalytic function of RNase E resides in A and processing of RNA molecules in Escherichia coli, the N-terminal half of the protein (amino acid residues an endoribonuclease, RNase E, has been shown to play 1–498), which also contains cleavage site specificity a major role in these processes. It is a multifunctional (McDowall et al. 1995). Smaller RNase E derivatives that degrades bulk RNA (Ono and that contain the first 395 amino acid residues show a Kuwano 1979), initiates the decay of a large fraction weak cleavage activity in vitro and further truncation of mRNA (for recent reviews, see Coburn and Mackie leads to loss of enzymatic activities (Caruthers et al. 1999; Steege 2000) and regulatory RNAs (Masse et al. 2006). A recent study of the structure of RNase E further 2003; Morita et al. 2005) by cleaving them at highly divides the catalytic domain into several subdomains: specific sites, and assists in the maturation of a variety of the RNase H, S1, 59 sensor, DNase I, Zn, and small catalytic RNAs, including 10Sa RNA (Lin-Chao et al. domains (Callaghan et al. 2005). The arginine-rich 1999), M1 RNA (Gurevitz et al. 1983), 5S rRNA (Ghora RNA-binding domain located between amino acids 580 and Apirion 1978), and 16S rRNA (Li et al. 1999; Wachi and 700 is similar to one found in many RNA-binding et al. 1999). proteins (Taraseviciene et al. 1995), and the C-terminal The essential 118-kDa protein encoded by rne con- third of the RNase E protein serves as a scaffold for the tains 1061 amino acids that can be partitioned into formation of a multicomponent ‘‘’’ com- three functionally distinct domains (Casare´gola et al. plex composed of the 39 polynucleotide phosphorylase (PNPase), the RNA helicase RhlB, and the glycolytic enolase (Carpousis et al. 1994; 1 These authors equally contributed to this work. Miczak et al. 1996; Py et al. 1996; Vanzo et al. 1998; Liou 2Corresponding author: 221 Hueksok-Dong, Donjak-Gu, Department of eroy Life Science, Chung-Ang University, Seoul, Republic of Korea, 156-756. et al. 2001; L et al. 2002). For a recent review, see E-mail: [email protected] Carpousis (2007). RNase E has additionally been

Genetics 179: 1871–1879 (August 2008) 1872 E. Shin et al. shown to be capable of interacting with poly(A) poly- Protein work: N-Rne or mutant N-Rne proteins were purified merase (Raynal and Carpousis 1999), ribosomal from KSL2000 cells harboring pNRNE4 or pNRNE4 con- alapos eng taining mutations and Western blot analyses of Rne, N-Rne, protein S1 (K et al. 1997; F et al. 2001), ee orita and N-Rne-NC were carried out as described previously (L RNA-binding protein Hfq (M et al. 2005), and the et al. 2002). Affinity purification of N-Rne protein typically protein inhibitors of RNase E activity, RraA and RraB yields .95% purity (supplemental Figure S2). To measure CD (Lee et al. 2003; Gao et al. 2006). However, the N- spectra of N-Rne and N-Rne-L385P proteins, purified m terminal half (amino acid residues 1–498) is sufficient proteins were stored in a buffer containing 20 m Na H2PO4 m for cell survival (Kido et al. 1996; Ow et al. 2000). (pH 7.5) and 200 m NaCl at a concentration of 0.5 mg/ml. To prepare total proteins from KSL2000 1 pACYC177 (no Although significant progress has been made in de- arabinose) or KSL2000 1 pNRNE4 or pNRNE4-NC, cultures termining the functional importance of RNase E in the were grown to middle log phase in the presence of 0.1% degradation and processing of RNA transcripts (for arabinose, harvested, washed twice with plain Luria–Bertani review, see Coburn and Mackie 1999; Steege 2000) (LB) medium, and reinoculated into LB medium containing and the crystal structure of RNase E has been resolved no arabinose (OD600 ¼ 0.1). They were further incubated for allaghan 150 min (OD600 ¼ 0.5) at 37° and 250 rpm and harvested for (C et al. 2005), there is still limited under- total protein preparation. standing of the amino acid residues and structural motifs In vitro cleavage of BR13: Synthesis of 59-end-labeled BR13 that mediate RNase E binding to and cleavage of specific and universally labeled p23 RNA, gel mobility assay, cleavage in vivo RNA substrates, its 59 / 39 quasi-processive mode assay, and Northern blot analysis were performed as described ee of enzyme action (Caruthers et al. 2006), and its 59-end previously (L et al. 2003). The RNA bands in the gel were ackie detected using a Packard Cyclone Phosphorimager and the dependence (M 1998). While studies of RNase E intensity of each band was quantitated using OptiQuant. variants have revealed some of this information (Diwa et al. 2002; Briegel et al. 2006), an intensive and sys- tematic search for RNase E loss-of-function mutants RESULTS containing amino acid substitutions in the catalytic do- main has not been done. To identify loss-of-function A screening strategy to identify functional residues in RNase E mutants, we developed a genetic system that the catalytic domain of RNase E: Genetic analysis of allows the introduction of random mutations into the RNase E has been hampered by the fact that it is an coding region of the catalytic domain, expression of the essential protein in E. coli. To circumvent this problem, we mutant RNase E proteins, and detection of mutant phe- utilized an E. coli strain (KSL2000) in which a chromo- notypes in cells complemented in trans to allow bacterial somal deletion in rne is complemented by a plasmid-born cell growth. Using this approach, we identified residues rne gene under the control of an arabinose-inducible in the catalytic domain important for ribonucleolytic promoter (pBAD-RNE) (Lee et al. 2002). Addition of activity. We report here the results of a systematic search 0.1% arabinose to cultures of KSL2000 induces the syn- for isolation and characterization of RNase E mutants thesis of full-length RNase E at wild-type levels and con- showing a loss-of-function phenotype. sequently supports survival and growth of this rne deletion mutant; KSL2000 cells are unable to form amura MATERIALS AND METHODS colonies in the absence of arabinose (T et al. 2006). A compatible ampicillin-resistance (Apr) plasmid Introduction of random mutations in the coding region of (pNRNE4) expressing the N-terminal 498 amino acids the catalytic domain of Rne: To construct pNRNE4 plasmid of RNase E with a hexahistidine tag at the C terminus (N- (Tamura et al. 2006) containing random mutations in the coding region of N-Rne, gel-purified error-prone PCR prod- Rne) under the control of the isopropyl-thiogalactoside ucts digested with NotI and XbaI were ligated into pNRNE4 (IPTG)-inducible lacUV5 promoter was introduced into plasmid that was digested with the same restriction . KSL2000 (Figure 1A) and the resulting transformants The DNA fragment encoding N-Rne was mutagenized by were able to grow optimally in the presence of 10–100 amplifying it using error-prone PCR as previously described mm IPTG (Figure 1B). Under these conditions, the (Kim et al. 2006). Primers used were Nrne 59 (59-GAATTGT GAGCGGATAAC-39) and Nrne 39 (59-CTACCATCGGCGC steady-state level of N-Rne protein is about four times TACGT-39). the normal level of full-length Rne, as determined by Isolation and analysis of noncomplementing N-Rne mutants: Western blot analysis using antibody against N-Rne, and KSL2000 cells harboring pNRNE4-mut, which has random is sufficient for optimal growth of the rne deletion mutations in the coding region of the catalytic domain of mutant as previously reported (Lee et al. 2002; Lee RNase E, were individually tested on LB–agar medium contain- ohen ing 1–1000 mm IPTG to identify their ability to support the and C 2003). No full-length RNase E protein was growth of KSL2000 cells expressing mutant N-Rne only. Three detected in N-Rne-complemented (Figure 1C). of the mutations isolated (I41N, A326T, and L385P) were To identify functional residues in the catalytic domain subcloned into pLAC-RNE1-DH by ligating the NotI–PmlI of RNase E, the DNA segment encoding amino acids fragment of pNRNE4 containing the mutations into the same 1–498 of Rne was amplified using error-prone PCR, sites in pLAC-RNE1-DH. Plasmid pLAC-RNE1-DH was con- structed by ligating the HindIII–SphI fragment of pFUS1500 ligated into pNRNE4 by replacing the wild-type copy of (McDowall et al. 1995) containing the coding region for the C- N-rne, and introduced by transformation into KSL2000; terminal half of Rne into the HindIII–XbaI sites in pNRNE4. transformants were individually tested for their ability Functional Residues in the Catalytic Domain of RNase E 1873

to support the growth of KSL2000 cells on LB–agar medium containing 10–1000 mm IPTG. MnCl2 (0.1 mm) was added to the PCR reaction to induce approximately one to three nucleotide substitutions per amplified copy, as has been previously determined by the random mutagenesis of a DNA fragment of similar size (1.5 kbp) encoding 16S rRNA (Kim et al. 2006). Identification of functional residues in the catalytic domain of RNase E: A total of 15,000 transformants harboring pNRNE4 containing random mutations in the coding region of N-Rne were screened for the loss of ability to support colony formation by KSL2000 cells in the presence of 10–1000 mm IPTG. Sixty-eight clones were obtained, and Western blot analysis using anti- bodies against RNase E showed that 12 of these expressed truncated proteins as a result of introduction of non- sense mutations (data not shown). Clones expressing truncated proteins were excluded from further analysis and the mutated residues of the rest of the clones were identified. As shown in Figure 2A, 18 clones contain a single-amino-acid substitution while the others contain two to three substitutions (not shown). The single- amino-acid substitutions cluster mainly in the DNase I, RNase H, and S1 subdomains and are positioned on the same surface of the protein that has been shown to bind and cleave RNA (Figure 2B) (Callaghan et al. 2005). The degree of conservation of the wild-type amino acid residues that were substituted in the noncomple- menting N-Rne (N-Rne-NC) mutants was analyzed by comparing the amino acid sequences of E. coli RNase E homologs found in other bacterial chromosomal DNA sequences in the NCBI database. The results show that the wild-type residues are nearly invariably conserved among RNase E homologs in phylogenetically diverse bacterial species (Table 1). Decay of RNA I by N-Rne-NC in vivo: Seven of the noncomplementing mutants harboring a single-amino- acid substitution were further characterized to deter- mine the basis of the inability of these mutants to Figure 1.—Properties of the genetic system. (A) A genetic complement a deficiency of wild-type N-Rne. KSL2000 system to isolate N-Rne mutants. A compatible ampicillin resis- cells expressing N-Rne-NC containing a single-amino- tance (Apr) plasmid (pNRNE4) expressing the N-terminal 498 acid substitution in the RNase H (I6T), S1 (I41N, G44D, amino acids of RNase E with a hexahistidine tag at the C termi- and T117I), or DNase I (A326T, I348T, and L385P) nus (N-Rne) harboring random amino acid substitutions under subdomain were conditionally expressed in the absence the control of the isopropyl-thiogalactoside (IPTG)-inducible lacUV5 promoter was introduced into KSL2000 in which a chro- of full-length RNase E to determine the ribonucleolytic mosomal deletion in rne was complemented by a plasmid-borne activity of the mutants in the cell. KSL2000 cells con- rne gene under the control of an arabinose-inducible BAD pro- ditionally depleted for Rne by transferring bacteria to moter (pBAD-RNE). (B) Growth characteristics of cells express- liquid media lacking arabinose underwent two to three ing noncomplementing N-Rne mutants. KSL2000 cells cell divisions at a doubling rate similar to that observed harboring pNRNE4 or pNRNE4-NC were individually tested on LB–agar medium containing 10–1000 mm IPTG for their for bacteria induced by 0.2% arabinose to express Rne ability to support the growth of KSL2000 cells. KSL2000 con- at endogenous levels (data not shown). This result is taining pACYC177 grew only when full-length RNase E was ex- consistent with the previous finding showing that E. coli pressed from pBAD-RNE. (C) Expression profiles of Rne and cell division requires a cellular RNase E concentration at N-Rne in KSL2000. The membrane probed with anti-Rne least 10–20% of normal ( Jain et al. 2002). monoclonal antibody was stripped and subsequently reprobed with anti-S1 polyclonal antibody to provide an internal stan- Using this characteristic of KSL2000, we analyzed the dard. The relative abundance of protein bands was quantitated steady-state level of a well-studied RNase E substrate in using the Versa Doc imaging system and Quantity One. KSL2000 cells conditionally expressing N-Rne-NC in 1874 E. Shin et al.

Figure 2.—Distribution of amino acid substi- tutions eliminating the ability of N-Rne to sup- port the growth of E. coli. (A) The N-terminal domain (residues 1–529) of RNase E is divided in- to subdomains as indicated. Isolated single-amino- acid substitutions eliminating N-Rne’s ability to support growth of E. coli are indicated above, re- spectively, the rectangle representing the N-terminal domain. (B) Isolated single-amino-acid substitu- tions are positioned in the crystal structure of the N-terminal domain of RNase E (Callaghan et al. 2005).

the absence of the full-length RNase E to determine Processing of essential noncoding RNAs in an rne- whether the inability of N-Rne mutants to support cell deleted strain expressing N-Rne-NC: The initial discov- viability results from the loss of cellular ribonucleolytic ery of RNase E was based upon its ability to process 9S activity. This RNase E substrate was RNA I, an antisense rRNA in E. coli cells and the finding that a shift of rne ts regulator of ColE1-type plasmid DNA replication; as bacteria to a nonpermissive temperature leads to the previously shown RNA I abundance controls the copy number of the plasmid (Lin-Chao et al. 1994) and this TABLE 1 function has been used to assess the ribonucleolytic activity of RNase E in vivo (Lee and Cohen 2003; Yeom Conservation of amino acid residues in RNase E-like proteins and Lee 2006). The induced expression of the N-Rne- Conservation NC mutants in the absence of wild-type RNase E in Amino acid KSL2000 cells resulted in a decrease in copy number of position Amino acid Identity Similarity the ColE1-type plasmid pNRNE4, which contains the 6 Ile 83/83 83/83 noncomplementing mutations (pNRNE4-NC), by three 14 Ile 82/83 83/83 to fourfold relative to that observed in KSL2000 cells 41 Ile 83/83 83/83 expressing wild-type N-Rne (Figure 3A). Western blot 44 Gly 83/83 83/83 analysis of N-Rne-NC proteins revealed that the amount 53 Leu 83/83 83/83 of N-Rne-NC proteins in these cells was 70–80% of 60 Tyr 71/83 83/83 wild-type N-Rne due to the decreased copy number of 68 Leu 83/83 83/83 the plasmid that expresses N-Rne-NC (Figure 3B). 73 Ile 81/83 83/83 112 Lys 83/83 83/83 These results show that N-Rne-NC mutants have lost 117 Thr 83/83 83/83 the ability to cleave RNA I molecules and consequently 232 Asp 83/83 83/83 have a lower copy number of the ColE1-type plasmid. 290 Ser 81/83 81/83 Changes in copy number in cells expressing N-Rne-NC 326 Ala 83/83 83/83 are likely to be amplified because pNRNE4 is a ColE1- 341 Gly 83/83 83/83 type plasmid; however, changes in copy number do 348 Ile 83/83 83/83 linearly reflect the extent of decrease in RNase E 385 Leu 83/83 83/83 439 Val 72/83 83/83 activity. Functional Residues in the Catalytic Domain of RNase E 1875

Figure 3.—Effect of N-Rne-NC on the stability of RNase E substrate RNAs in vivo. (A) Decay of RNA I. KSL2000 cells harboring pNRNE4 or pNRNE4-NC were depleted for Rne as described in materials and methods for plasmid preparation. Plasmids di- gested with HindIII , which has a unique cleavage site in all plasmids tested here, were electrophoresed in a 0.9% agarose gel and stained with ethidium bromide. Plasmid copy numbers were calculated relative to a concurrently present pSC101 derivative (pBAD-RNE), the replication of which is independent of Rne, by measuring the molar ratio of the pBAD-RNE plasmid to ColE1-type plasmid (pNRNE4 or pNRNE-NC), and are shown at the bottom of the gel. (B) Expression profiles of N-Rne and N-Rne-NC in KSL2000 conditionally depleted for Rne. The same procedure described in the Figure 1C legend was used for Western blot analysis. Processing of pM1 (C), tRNAAsn (D), and 9S (E) is shown. Total RNA was separated in a 6% (pM1) or an 8% (tRNAAsn and 9S) PAGE containing 8 m urea. Separated RNA bands were transferred to a Nylon membrane and probed with 32P end-labeled oligos complementary to each RNA molecule. Percentages of precursors were calculated as previously described (Lee and Cohen 2003). KSL2000 cells expressing N-Rne-NC were grown for plasmid and total RNA preparation as described in materials and methods. This procedure was applied to remove the full-length RNase E expressed from pBAD-RNE and to mea- sure the ribonucleolytic activity of N-Rne mutants in the cell. in vivo accumulation of precursors of 5S rRNA (Ghora rne-deleted cells in which synthesis of RNase E from the and Apirion 1978), pM1 RNA (Reed et al. 1982), and pBAD-RNE plasmid (i.e., the KSL2000 strain) was tRNAAsn (Li and Deutscher 2002; Ow and Kushner turned off by shift to media lacking arabinose (Figure 2002). To test the ability of N-Rne-NC mutants to pro- 3, C–E, lane 1). In contrast, the RNAs were processed cess precursors of these essential noncoding RNA tran- normally in KSL2000 cells expressing wild-type N-Rne scripts, we measured the steady-state transcript levels in (Figure 3, C–E, lane 3). These results suggest that these cells conditionally expressing N-Rne-NC mutants con- N-Rne-NC mutants have little to no ribonucleolytic taining a single-amino-acid substitution in the S1 sub- activity in vivo. domain (I41N), which previously has been shown to be Effects of noncomplementing single-amino-acid sub- implicated in the binding of RNase E to substrate RNA stitutions on the ribonucleolytic activity of full-length (Schubert et al. 2004; Callaghan et al. 2005), and in RNase E: To confirm that the loss of ribonucleolytic the DNase I subdomain (A326T and L385P), which is activity by substitution of a single amino acid in these spatially distant from the catalytic and RNA- mutant proteins is not a property of only the truncated of the protein (Callaghan et al. 2005). We found that N-Rne protein, three of the mutations (I41N, A326T, Rne-depleted cells expressing these N-Rne-NC mutants and L385P) were subcloned into a plasmid expressing were deficient in the processing of precursors of all of full-length RNase E under the control of the IPTG- these RNAs (Figure 3, C–E). The precursor bands inducible lacUV5 promoter (pLAC-RNE1-DH) and tested accumulating in rne-deleted cells expressing the non- for their ability to support the growth of KSL2000 cells in complementing N-Rne mutants were identical in size to, thepresenceof1–1000mm IPTG. While KSL2000 cells and similar in quantity to, the species accumulating in expressing wild-type RNase E from pLAC-RNE1-DH 1876 E. Shin et al. grew normally in the presence of 1–10 mm IPTG, none of the RNase E mutants containing a noncomplementing mutation supported the growth of KSL2000 at IPTG concentrations of 1–1000 mm, indicating that the effects of these amino acid substitutions are not specific to N- Rne (data not shown). In vitro cleavage activity of N-Rne-NC: To test whether the observed evidence of decreased ribonu- cleolytic activity of N-Rne-NC mutants in vivo resulted from defective ribonucleolytic activity of the mutated N-Rne-NC proteins, we measured the in vitro cleavage rates of wild-type and N-Rne-NC proteins on BR13, an oligoribonucleotide that contains the RNase E-cleaved sequence of RNA I. Affinity-purified wild-type or mutant N-Rne proteins containing a single-amino-acid substitu- tion (I41N, A326T, and L385P) were incubated with 59-end-labeled BR13. None of the three N-Rne-NC proteins tested detectably cleaved BR13, whereas this oligoribonucleotide was cleaved efficiently by the wild- type N-Rne protein (Figure 4A). Characterization of N-Rne-NC proteins bearing mutations in the region of DNase I subdomain spatially distant from the catalytic site: We were particularly interested in learning the basis for the observed loss of ribonucleolytic function for mutants containing single- amino-acid substitutions within the DNase I subdomain, which is not in close proximity to the site implicated by crystallographic analysis of the N-Rne BR13 complex (Figure 2B) in the binding and cleavage of BR13 (Callaghan et al. 2005). Although the functional role of this region has not been previously identified, mutations in the region could in principle interfere Figure 4.—(A) Cleavage of BR13 by N-Rne and N-Rne-NC with the ribonucleolytic activity of RNase E by either in vitro. One-half picomole of 59-end-labeled BR13 was incu- reducing binding to the RNA substrate or inhibiting the bated with 50 ng of purified N-Rne or N-Rne-NC in cleavage catalytic activity of the enzyme. To further understand buffer at 37°. Each sample was removed at each time point the basis for the loss of ribonucleolytic activity of these indicated and mixed with an equal volume of loading buffer. mutant proteins, we tested the ability of N-Rne-NC Samples were denatured at 75° for 5 min and loaded onto 15% polyacrylamide gels containing 8 m urea. The radioactiv- proteins mutated in the DNase I subdomain (A326T ity in each band was quantitated using phosphorimager and and L385P) to bind to p23 RNA, which is a truncated OptiQuant. (B) RNA-binding activity of N-Rne-NC. One-half pM1 RNA that is processed by RNase E to a product picomole of internally labeled p23 RNA was incubated at termed 23 RNA (Kim et al. 1996). Gel shift assays room temperature for 10 min with 50 ng of proteins indicated m 3 indicated the inability of these N-Rne-NC mutants to in 20 lof1 cleavage buffer. After detecting RNA bands us- ing phosphorimager, proteins in the gel was transferred to a bind to p23 RNA (Figure 4B). We detected the major nitrocellulose membrane and probed with monoclonal anti- band (band b in Figure 4B) that did not interact with bodies to Rne. wild-type protein and consequently did not shift. We think that it is probably p23 RNA molecules with a dif- ferent structure that does not allow wild-type N-Rne to when an untruncated version of p23 RNA (pM1) was bind. Even though the in vitro synthesized p23 RNA was used for the gel mobility shift assay (Lee et al. 2003). denatured at 75° and renatured by slowly cooling down In contrast to wild-type N-Rne, the N-Rne-A326T and to room temperature, two separate RNA bands (bands a N-Rne-L385P mutants in the DNase I subdomain as well and b) were formed in the gel. It is possible that the as the N-Rne-I41N mutant in the S1 subdomain all complex formed between wild-type protein and p23 showed no detectable binding to p23 RNA under the RNA present in band a is more stable than the one same conditions, suggesting that the defective ribonu- formed between the protein and p23 RNA present in cleolytic activity of N-Rne-A326T and N-Rne-L385P re- band b under the conditions used for the gel mobility sults from reduction in the substrate-binding ability of shift assay. A similar phenomenon has been observed the enzyme. When higher concentrations of proteins were Functional Residues in the Catalytic Domain of RNase E 1877

Figure 5.—Cleavage ac- tivity of N-Rne-NC. (A) In vitro cleavage of p23 RNA by N-Rne-NC. Each protein (wild type, I41N, A326T, or L385P) at the concentration of 20 nm was incubated with various con- centrations of p23 (18.8– 600 nm) for 15 min at 37° and analyzed in an 8% PAGE containing 8 m urea. (B) Quantitation of cleavage activity of N-Rne-NC. The ra- dioactivity in each band was quantitated using phosphor- imager and OptiQuant and plotted. (C) Detection of misfolding of N-Rne-L385P. Purified proteins of N-Rne and N-Rne-L385P were used tomeasuretheCDspectrum.

used for the gel mobility shift assay, p23 RNA incubated cently resolved crystal structure of N-terminal RNase E with wild-type N-Rne was cleaved and consequently the complexed with a short (10–15 nt) RNA oligonucleo- uncleaved RNA bands (bands a and b in Figure 4B, lane 2) tide having 29-O-methyl modifications (Callaghan et al. were converted to new bands below band a (lane 2 in 2005) did not overlap with the noncomplementing supplemental Figure S1). However, including higher residues isolated in this study except for the lysine levels of mutant protein N-Rne-L385P did not result in a residue at position 112. The functionally important res- shift of any bands (lane 3 in supplemental Figure S1). idues identified by Callaghan et al. (2005) are all im- Consistent with this observation, at the highest sub- plicated in engagement of the terminal phosphate strate concentrations tested, cleavage products were (G124, V128, R169, T170, and R373), forming a hy- observed for the N-Rne-A326T and N-Rne-L385P pro- drophobic pocket that binds the nucleotide adjacent to teins, although the yield was much less than that for the the cleavage site (F57, F67, and K112), nucleophilic wild-type enzyme (Figure 5, A and B). However, N-Rne- attack of the scissile phosphate (D303, N305, and I41N, which failed to show any binding activity (Figure D346), and contact with the exocyclic oxygen of the 4), showed no cleavage activity at any substrate concen- base at position 1 with regard to the scissile phosphate tration tested. As proline substitutions commonly lead (K106). As selective replacement by Callaghan et al. to disruption of protein structure, we compared the (2005) of functionally important RNase E residues with structure of N-Rne-L385P protein with wild-type N-Rne, other amino acids based on the crystal structure re- using circular dichroism (CD) to learn whether the sulted in the loss of RNA-binding ability, RNA cleavage mutation that abolishes the binding and enzymatic ac- activity, or both in vitro, it was surprising to find only one tivity of the N-Rne-L385P protein leads to misfolding of overlapping position (K112E) among mutations that the protein. As shown in Figure 5C, the CD spectrum of prevented RNase E from supporting cell viability. One of N-Rne-L385P protein was nearly identical to that of several possibilities may account for this absence of wild-type N-Rne, indicating that there is no significant overlap. Although we screened 15,000 clones to isolate collapse or misfolding of the mutant protein. 18 noncomplementing mutants bearing single-amino- acid substitutions, we observed a low frequency of mu- tation redundancy, implying that the library of possible DISCUSSION mutations was not fully saturated. Additionally, the The isolated single-amino-acid substitutions eliminat- amino acids chosen for mutagenesis based on the crystal ing the ability of RNase E to support survival of E. coli structure were all implicated in contacts with small cells are clustered in the S1, DNase I, and RNase H oligonucleotides, which are unlikely to be major RNase subdomains and positioned on the same face of the E substrates in vivo and, therefore, might not represent protein as the RNA-binding and cleavage sites. However, all functional residues of RNase E that interact with 11 functionally important residues identified in a re- in vivo RNA substrates, which are much longer and more 1878 E. Shin et al. complex than small oligonucleotides. It is also possible subdomain that contains these amino acid substitutions that some of the residues found in our mutants are not and that is spatially distant from the catalytic site of directly involved in binding or cleaving RNA, but rather RNase E may contain additional RNA-binding sequen- in forming structural elements to maintain the active ces or may modulate the binding of other enzyme re- form of the enzyme. One such example is the mutation gions to substrates. It also remains possible that L68P recovered in our screen that had been previously misfolding of mutant proteins to a degree that was not identified in two conditional mutants (G66S and L68F) detected by CD analysis is the basis for the phenotype. that lead to a lethal phenotype at elevated temperatures; However, any such misfolding would likely be limited these mutations were inferred to globally destabilize the and localized, as some RNA cleavage activity is retained folded structure of the RNase E S1 domain (Schubert by the mutated protein. Rather, it is tempting to spec- et al. 2004). A final possibility is that the residues we ulate that wild-type residues at these mutated positions identified mediate enzyme functions that cannot be in the DNase I subdomain may constitute a highly con- inferred from the crystal structure. For example, it has served structural motif that aids RNase E-like enzymes in been proposed that multimeric forms of RNase E can be the selection of specific cleavage sites in AU-rich, single- catalytically activated by the allosteric effector 59 mono- stranded regions. This motif may facilitate enzymatic phosphate present in target RNAs, which induces sig- attack on such regions by overcoming impediments nificant structural changes in the protein that both imposed by cis- and/or trans-acting elements such as enhance catalytic activity and constrain substrate binding high-order RNA structures and RNA-binding proteins (Jiang and Belasco 2004), and the enzyme may require present in the vicinity of the cleavage sites. It is unlikely such augmented binding and catalytic activity to cleave that all amino acid substitutions identified in the some substrates in vivo that are long structured RNAs. DNase I subdomain (Figure 2A) are implicated in The wild-type amino acid residues at the mutated forming the additional RNA-binding motif since the positions we identified, which are mainly clustered in the region proximal to the catalytic site of RNase E that S1, DNase I, and RNase H subdomains, are nearly invari- includes D303, N305, and D346 has been proposed to ably conserved in RNase E homologs, consistent with our interact with the hydrated magnesium ion, the activated finding that they are essential for cell survival. However, water molecule that attacks the scissile phosphate of the notwithstanding the essentiality of normal residues at RNA (Callaghan et al. 2005). these positions, two of the three the mutant proteins we This research was supported by grants from the 21C Frontier tested bind to RNA (albeit poorly) and show in vitro Microbial Genomics and Application Center Program of the Korean ribonucleolytic activity at high substrate concentrations. Ministry of Science & Technology and the Basic Research Program of The ability of RNase E-like enzymes to cleave AU-rich the Korea Science and Engineering Foundation (R01-2005-000-10293-0) single-stranded regions of numerous target RNA mole- to K.L. and Y.L. and from the National Institutes of Health (AI08619) to S.N.C. cules implies that these proteins have a conserved structural motif for selecting cleavage sites present in structurally complex RNA substrates. As already noted, LITERATURE CITED the only residue so far identified on the basis of the Briegel, K. J., A. Baker and C. Jain, 2006 Identification and anal- crystal structure for sequence recognition is K106, and it ysis of Escherichia coli ribonuclease E dominant-negative mutants. Genetics 172: 7–15. has been inferred that the preference of RNase E for Callaghan, A. J., M. J. Marcaida,J.A.Stead,K.J.McDowall,W.G. AU-rich substrates results mainly through the recogni- Scott et al., 2005 Structure of Escherichia coli RNase E catalytic do- tion of the RNA conformation (Callaghan et al. 2005). main and implications for RNA turnover. Nature 437: 1187–1191. Carpousis, A. J., 2007 The RNA degradosome of Escherichia coli:an However, it is known that the enzyme does not simply mRNA-degrading machine assembled on RNase E. Annu. Rev. cleave AU-rich single-stranded regions of RNA mole- Microbiol. 61: 71–87. cules but rather cleaves RNA sequences with high Carpousis, A. J., G. V. Houwe,C.Ehretsmann and H. M. Krisch, in hao c owall 1994 Copurification of E. coli RNase E and PNPase: evidence specificity (L -C et al. 1994; M D et al. for a specific association between two enzymes important in 1995; Kaberdin 2003). Moreover, in vivo RNase E sub- RNA processing and degradation. Cell 76: 889–900. strates have cleavage sites that commonly are preceded Casare´gola, S., A. Jacq,D.Laoudj,G.McGurk,S.Margarson et al., 1992 Cloning and analysis of the entire Escherichia coli or followed by a stem-loop structure that seems to ams gene: ams is identical to hmp1 and encodes a 114 kDa pro- modulate the degradation rate of the RNA transcripts tein that migrates as a 180 kDa protein. J. Mol. Biol. 228: 30–40. (Melefors and Von Gabain 1988; Lin-Chao and Caruthers,J.M.,Y.Feng,D.B.McKay andS.N.Cohen, ohen ormack ackie ackie 2006 Retention of core catalytic functions by a conserved mini- C 1991; C and M 1992; M and mal ribonuclease E peptide that lacks the domain required for tet- Genereaux 1993; Diwa et al. 2000). Therefore, it is ramer formation. J. Biol. Chem. 281: 27046–27051. likely that the RNase E protein has additional RNA- Coburn, G. A., and G. A. Mackie, 1999 Degradation of mRNA in Escherichia coli: an old problem with some new twists. Prog. Nu- binding motifs to select AU-rich single-stranded regions cleic Acid Res. Mol. Biol. 62: 55–108. in highly structured RNA substrates. Our finding that Cormack, R. S., and G. A. Mackie, 1992 Structural requirements for mutant proteins bearing amino acid substitutions the processing of Escherichia coli 5 S ribosomal RNA by RNase E in vitro. J. Mol. Biol. 228: 1078–1090. A326T and L385P fail to bind RNA but still retain Diwa, A., A. L. Bricker,C.Jain and J. G. Belasco, 2000 An evolu- catalytic activity suggests that the region of the DNase I tionarily conserved RNA stem-loop functions as a sensor that di- Functional Residues in the Catalytic Domain of RNase E 1879

rects feedback regulation of RNase E gene expression. Genes ticomponent complexes associated with the cytoplasmic mem- Dev. 14: 1249–1260. brane via the N-terminal region of ribonuclease E. Proc. Natl. Diwa, A. A., X. Jiang,M.Schapira and J. G. Belasco, 2002 Two Acad. Sci. USA 98: 63–68. distinct regions on the surface of an RNA-binding domain are Mackie, G. A., 1998 Ribonuclease E is a 59-end-dependent endonu- crucial for RNase E function. Mol. Microbiol. 46: 959–969. clease. Nature 395: 720–723. Feng, Y., H. Huang,J.Liao and S. N. Cohen, 2001 Escherichia coli Mackie, G. A., and J. L. Genereaux, 1993 The role of RNA structure poly(A)-binding proteins that interact with components of degra- in determining RNase E-dependent cleavage sites in the mRNA for dosomes or impede RNA decay mediated by polynucleotide ribosomal protein S20 in vitro. J. Mol. Biol. 234: 998–1012. phosphorylase and RNase E. J. Biol. Chem. 276: 31651–31656. Masse, E., F. E. Escorcia and S. Gottesman, 2003 Coupled degra- Gao, J., K. Lee,M.Zhao,J.Qiu,X.Zhan et al., 2006 Differential dation of a small regulatory RNA and its mRNA targets in Escher- modulation of E. coli mRNA abundance by inhibitory proteins ichia coli. Genes Dev. 17: 2374–2383. that alter the composition of the degradosome. Mol. Microbiol. McDowall, K. J., V. R. Kaberdin,S.W.Wu,S.N.Cohen and S. Lin- 61: 394–406. Chao, 1995 Site-specific RNase E cleavage of oligonucleotides Ghora, B. K., and D. Apirion, 1978 Structural analysis and in vitro and inhibition by stem-loops. Nature 374: 287–290. processing to p5 rRNA of a 9S RNA molecule isolated from an rne Melefors, O., and A. von Gabain, 1988 Site-specific endonucleo- mutant of E. coli. Cell 15: 1055–1066. lytic cleavages and the regulation of stability of E. coli ompA Gurevitz, M., S. K. Jain and D. Apirion, 1983 Identification of a mRNA. Cell 52: 893–901. precursor molecule for the RNA moiety of the processing en- Miczak, A., V. R. Kaberdin, C.-L. Wei and S. Lin-Chao, 1996 Pro- zyme RNase P. Proc. Natl. Acad. Sci. USA 80: 4450–4454. teins associated with RNase E in a multicomponent ribonucleo- Jain, C., A. Deana and J. G. Belasco, 2002 Consequences of RNase lytic complex. Proc. Natl. Acad. Sci. USA 93: 3865–3869. E scarcity in Escherichia coli. Mol. Microbiol. 43: 1053–1064. Morita, T., K. Maki and H. Aiba, 2005 RNase E-based ribonucleo- Jiang, X., and J. G. Belasco, 2004 Catalytic activation of multimeric protein complexes: mechanical basis of mRNA destabilization RNase E and RNase G by 59-monophosphorylated RNA. Proc. mediated by bacterial noncoding RNAs. Genes Dev. 19: 2176– Natl. Acad. Sci. USA 101: 9211–9216. 2186. Kaberdin, V. R., 2003 Probing the substrate specificity of Escherichia Ono, M., and M. Kuwano, 1979 A conditional lethal mutation in coli RNase E using a novel oligonucleotide-based assay. Nucleic Escherichia coli strain with a longer chemical lifetime of messenger Acids Res. 31: 4710–4716. RNA. J. Mol. Biol. 129: 343–357. Kalapos, M. P., H. Paulusb and N. Sarkara, 1997 Identification of Ow, M. C., and S. R. Kushner, 2002 Initiation of tRNA maturation ribosomal protein S1 as a poly(A) binding protein in Escherichia by RNase E is essential for cell viability in E. coli. Genes Dev. 16: coli. Biochimie 79: 493–502. 1102–1115. Kido,M.,K. Yamanaka,T.Mitani,H.Niki,T.Oguraetal., 1996 RNase Ow, M. C., Q. Liu and S. R. Kushner, 2000 Analysis of mRNA decay E polypeptides lacking a carboxyl-terminal half suppress a mukB and rRNA processing in Escherichia coli in the absence of RNase E- mutation in Escherichia coli. J. Bacteriol. 178: 3917–3925. based degradosome assembly. Mol. Microbiol. 38: 854–866. Kim, J. M., H. Go,W.S.Song,S.M.Ryou and K. Lee, 2006 Func- Py, B., C. F. Higgins,H.M.Krisch and A. J. Carpousis, 1996 A tional analysis of the residue 789 in Escherichia coli 16S rRNA and DEADbox RNA helicase in the Escherichia coli RNA degradosome. development of a method to select second-site revertants. Korean Nature 381: 169–172. J. Microbiol. 42: 156–159. Raynal, L. C., and A. J. Carpousis, 1999 Poly(A) polymerase I of Kim, S., H. Kim,I.Park and Y. Lee, 1996 Mutational analysis of RNA Escherichia coli: characterization of the catalytic domain, an structures and sequences postulated to affect 39 processing of M1 RNA binding site and regions for the interaction with proteins RNA, the RNA component of Escherichia coli RNase P. J. Biol. involved in mRNA degradation. Mol. Microbiol. 32: 765–775. Chem. 271: 19330–19337. Reed, R. E., M. F. Baer,C.Guerrier-Takada,H.Donis-Keller and S. Lee, K., and S. N. Cohen, 2003 A Streptomyces coelicolor functional or- Altman, 1982 Nucleotide sequence of the gene encoding the thologue of Escherichia coli RNase E shows shuffling of catalytic RNA subunit (M1 RNA) of from Escherichia coli. and PNPase-binding domains. Mol. Microbiol. 48: 349–360. Cell 30: 627–636. Lee, K., J. A. Bernstein and S. N. Cohen, 2002 RNase G comple- Schubert, M., R. E. Edge,P.Lario,M.A.Cook,N.C.Strynadka mentation of rne null mutation identifies functional interrela- et al., 2004 Structural characterization of the RNase E S1 do- tionships with RNase E in Escherichia coli. Mol. Microbiol. 43: main and identification of its oligonucleotide-binding and di- 1445–1456. merization interfaces. J. Mol. Biol. 30: 37–54. Lee, K., X. Zhan,J.Gao,J.Qiu,Y.Feng et al., 2003 RraA: a protein Steege, D. A., 2000 Emerging features of mRNA decay in bacteria. inhibitor of RNase E activity that globally modulates RNA abun- RNA 6: 1079–1090. dance in E. coli. Cell 114: 623–634. Tamura, M., K. Lee,C.A.Miller,C.J.Moore,Y.Shirako et al., Leroy, A., N. F. Vanzo,S.Sousa,M.Dreyfus and A. J. Carpousis, 2006 RNase E maintenance of proper FtsZ/FtsA ratio required 2002 Function in Escherichia coli of the non-catalytic part of RN- for nonfilamentous growth of Escherichia coli cells but not for col- ase E: role in the degradation of ribosome-free mRNA. Mol. Mi- ony-forming ability. J. Bacteriol. 188: 5145–5152. crobiol. 45: 1231–1243. Taraseviciene, L., G. R. Bjork and B. E. Uhlin, 1995 Evidence for Li, Z., and M. P. Deutscher, 2002 RNase E plays an essential role in an RNA binding region in the Escherichia coli processing endori- the maturation of Escherichia coli tRNA precursors. RNA 8: 97–109. bonuclease RNase E. J. Biol. Chem. 270: 26391–26398. Li, Z., S. Pandit and M. P. Deutscher, 1999 RNase G (CafA pro- Vanzo, N. F., Y. S. Li,B.Py,E.Blum,C.F.Higgins et al., tein) and RNase E are both required for the 59 maturation of 1998 Ribonuclease E organizes the protein interactions in 16S ribosomal RNA. EMBO J. 18: 2878–2885. the Escherichia coli RNA degradosome. Genes Dev. 12: 2770–2781. Lin-Chao, S., and S. N. Cohen, 1991 The rate of processing and Wachi, M., G. Umitsuki,M.Shimizu,A.Takada and K. Nagai, degradation of antisense RNAI regulates the replication of 1999 Escherichia coli cafA gene encodes a novel RNase, desig- ColE1-type plasmids in vivo. Cell 65: 1233–1242. nated as RNase G, involved in processing of the 59 end of 16S Lin-Chao, S., T. T. Wong,K.J.McDowall and S. N. Cohen, rRNA. Biochem. Biophys. Res. Commun. 259: 483–488. 1994 Effects of nucleotide sequence on the specificity of rne- Yeom, J.-H., and K. Lee, 2006 RraA rescues Escherichia coli cells over- dependent and RNase E-mediated cleavages of RNA I encoded producing RNase E from growth arrest by modulating the ribo- by the pBR322 plasmid. J. Biol. Chem. 269: 10797–10803. nucleolytic activity. Biochem. Biophys. Res. Commun. 345: 1372– Lin-Chao, S., C. L. Wei and Y. T. Lin, 1999 RNase E is required for 1376. the maturation of ssrA RNA and normal ssrA RNA peptide- tagging activity. Proc. Natl. Acad. Sci. USA 96: 12406–12411. Liou, G.-G., W.-N. Jane,S.N.Cohen, N.-S. Lin and S. Lin-Chao, 2001 RNA degradosomes exist in vivo in Escherichia coli as mul- Communicating editor: S. Gottesman